Message-ID: <1156d735-5a1b-4c5e-8e13-44b15ec1ee21@postgrespro.ru>
Date: Tue, 17 Oct 2023 10:11:12 +0300
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: remaining sql/json patches
Content-Language: en-US
To: Amit Langote <amitlangote09@gmail.com>
Cc: jian he <jian.universality@gmail.com>, Nikita Malakhov
 <hukutoc@gmail.com>, Andres Freund <andres@anarazel.de>,
 Alvaro Herrera <alvherre@alvh.no-ip.org>,
 Andrew Dunstan <andrew@dunslane.net>, Erik Rijkers <er@xs4all.nl>,
 PostgreSQL-development <pgsql-hackers@postgresql.org>
References: 
 <CA+HiwqFJQdBQzTbSt6=S_3mmYf_pb2YoRNrhL1R0zzFz5Yja=w@mail.gmail.com>
 <202309281104.kmkr4v34ny2t@alvherre.pgsql>
 <CA+HiwqETtgH0ESDEEGDwTmazmtNeR7UsJS=BOoa=ghdYMo+Ufg@mail.gmail.com>
 <20231006214856.yie5dedhzrcb3qxx@alap3.anarazel.de>
 <CA+HiwqGdrSTsjyBcw66WE=vDke4B5fJj2zXBbXpv-npgzm4+rg@mail.gmail.com>
 <CA+HiwqEHWkGwiy3mKzjf0MaPU84LZh9PjtL3TDZuvF=0QLVz=w@mail.gmail.com>
 <CAN-LCVNNjsruFkAGHeF-9W0yTQKuELKAXsyYs3WDTe_SX+9ppg@mail.gmail.com>
 <CAN-LCVP=5QSmbndz4q+mNxddLa2y4XVOw+z7Dx=iUkegNM8Sjg@mail.gmail.com>
 <CA+HiwqGNmJ67n6BKAROEBHqv0faGA3GxU+vH1wO8x-m9i=hRoQ@mail.gmail.com>
 <CACJufxHLYQPw-7Wg=EYqnQPjhaB_dJGH+USsT0vP5FL3sxLX5Q@mail.gmail.com>
 <c04ff761-930b-4345-ae23-f78b8ee7ce32@postgrespro.ru>
 <CA+HiwqFcrOVYXzV9pRME+p-B6cTQM5XDBfws+_0rqjYg9k4kLA@mail.gmail.com>
From: "Anton A. Melnikov" <a.melnikov@postgrespro.ru>
In-Reply-To: 
 <CA+HiwqFcrOVYXzV9pRME+p-B6cTQM5XDBfws+_0rqjYg9k4kLA@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Archived-At: 
 <https://www.postgresql.org/message-id/1156d735-5a1b-4c5e-8e13-44b15ec1ee21%40postgrespro.ru>
Precedence: bulk


On 17.10.2023 07:02, Amit Langote wrote:

> One thing jian he missed during the debugging is that
> ExecEvalJsonExprCoersion() receives the EMPTY ARRAY value via
> *op->resvalue/resnull, set by ExecEvalJsonExprBehavior(), because
> that's the ON EMPTY behavior specified in the constraint.  The bug was
> that the code in ExecEvalJsonExprCoercion() failed to set val_string
> to that value ("[]") before passing to InputFunctionCallSafe(), so the
> latter would assume the input is NULL.
>
Thank a lot for this remark!

I tried to dig to the transformJsonOutput() to fix it earlier at the analyze stage,
but it looks like a rather hard way.

Maybe simple in accordance with you note remove the second condition from this line:
if (jb && JB_ROOT_IS_SCALAR(jb)) ?

There is a simplified reproduction before such a fix:
postgres=# select JSON_QUERY(jsonb '[]', '$' RETURNING char(5) OMIT QUOTES EMPTY ON EMPTY);
server closed the connection unexpectedly
         This probably means the server terminated abnormally

after:
postgres=# select JSON_QUERY(jsonb '[]', '$' RETURNING char(5) OMIT QUOTES EMPTY ON EMPTY);
  json_query
------------
  []
(1 row)

And at the moment i havn't found any side effects of that fix.
Please point me if i'm missing something.

With the best wishes!

-- 
Anton A. Melnikov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company